CN106331678A

CN106331678A - Hardware-implemented floating-point type viewpoint processing method in glasses-free 3D display system

Info

Publication number: CN106331678A
Application number: CN201610833722.4A
Authority: CN
Inventors: 秦剑飞
Original assignee: Sichuan Changhong Electric Co Ltd
Current assignee: Sichuan Changhong Electric Co Ltd
Priority date: 2016-09-20
Filing date: 2016-09-20
Publication date: 2017-01-11
Anticipated expiration: 2036-09-20
Also published as: WO2018054173A1; CN106331678B

Abstract

The invention relates to a glasses-free 3D display technology, discloses a hardware-implemented floating-point type viewpoint processing method in a glasses-free 3D display system, and aims to perform floating point processing and error compensation with high efficiency and high precision under the condition of hardware implementation so as to satisfy the display effects of different glasses-free 3D screens when the grating tile angle and tile distance change. According to the invention, the viewpoint value of a first sub pixel in each row is calculated precisely; then, the integer parts and floating point parts of the viewpoint values of other sub pixels are respectively arranged and estimated, the estimation result is compensated, and a cycle even distribution law is adopted in estimation and compensation; and finally, the final viewpoint values are calculated based on the integer parts and compensated floating point parts of the viewpoint values of the sub pixels.

Description

Can hard-wired floating type viewpoint processing method in bore hole 3D display system

Technical field

The present invention relates to bore hole 3D Display Technique, be specifically related in bore hole 3D display system to regard by hard-wired floating type Point processing method.

Background technology

In recent years, ordinary consumer improves year by year for bore hole 3D TV demand, and therefore bore hole 3D television cost is constantly While ground reduces, also begin to the bore hole 3D television system that various implementation different editions occurs.From full SOC system to entirely GPU system, the most is that full GPU system realizes.In order to reduce cost, application A8 chip adds the side that simple FPGA controls Method, but the picture display effect obtained is less desirable, and cost minimizing is also far smaller than expected value.

Bore hole processing scheme on existing market can be divided into integer and floating type, integer scheme to lose major part signal, draws Face display effect is very poor；The complexity realized due to algorithm of floating type and amount of calculation are too big, the most only it Middle very small part realizes for hardware, does not also have a whole set of hard-wired method.

If a whole set of bore hole 3D system whole flow processs after rgb signal enters all are realized with hardware, bore hole 3D electricity To be substantially reduced depending on the cost of machine, average family is bought bore hole 3D TV and will be possibly realized.

Summary of the invention

The technical problem to be solved is: for the defect of prior art, proposes a kind of bore hole 3D display system In can hard-wired floating type viewpoint processing method, it is intended to efficient and high-precision float in the case of hardware is attainable Point processes and error compensation, with the satisfied different bore hole 3D screens display effect when grating slope angle and slant range change.

The technical solution adopted for the present invention to solve the technical problems is:

In bore hole 3D display system can hard-wired floating type viewpoint processing method, comprise the following steps:

A, parameter according to bore hole 3D screen, the viewpoint value of first sub-pixel of a line every to picture calculates；

B, viewpoint value N to first sub-pixel of every a line are cast out process and are obtained its integer part；

C, integer part to the N of the sub-pixel position of every a line are estimated；

D, floating-point part to the N of the sub-pixel position of every a line are averaged arranged evenly and carry out floating-point weight and estimate Calculate；

E, the error bringing viewpoint estimation carry out the backward compensation with the regularity of distribution；

F, according to estimation viewpoint integer part and compensate after floating-point weight portion calculate final viewpoint value.

As optimizing further, in step A, to the method for the viewpoint value calculating of first sub-pixel of every a line it is:

According to seeking viewpoint formula: N=M_out/P*V_viewCalculate viewpoint value, the viewpoint value asked for is retained eight floating-points； Wherein, N is viewpoint value, M_outMove up and down the computational constant of relation for display screen physical resolution and grating slope, P is grating Physical parameter, V_viewNumber for viewpoint.

As optimizing further, in step C, the integer part of the N of the sub-pixel position of every a line is arranged in order and by Estimate according to the rule moved in circles.

As optimizing further, in step D, the method for described floating-point weight estimation is:

Starting estimation from second sub-pixel of every a line, floating-point is increased by sub-pixel position, and increments is to zero, then From zero increase to this journey, the value of first sub-pixel, so circulates, until the floating-point weight of last sub-pixel position of this journey Estimation completes.

As optimizing further, in step E, the described error bringing viewpoint estimation carries out the backward benefit with the regularity of distribution Repay, specifically include:

The floating-point part cast out is needed to be defined as F in addition to eight floating numbers retained the viewpoint value asked in step A, will The sub-pixel of co-located, according to the mode of distribution same with the mode that the is evenly distributed backward in step D, is floated by the F of first sub-pixel Point weight compensates, and obtaining final floating-point part is N_F。

Floating-point weight portion as optimizing further, in step F, after the described viewpoint integer part according to estimation and compensation Point calculate final viewpoint value method be:

By formula N_Eventually=N_Z*N_F+(N_Z+1)*(1-N_F) calculate the viewpoint value that final upper screen multiple views merges.

The invention has the beneficial effects as follows: improve the efficiency of algorithm, floating type bore hole 3D can be realized with low cost hardware and show Show system, cost-effective.

Detailed description of the invention

Bore hole 3D algorithm in the present invention, from the beginning of the image data of HDMI or LVDS path inputs, each Step all hardware realizes, and utilizes DDR3 storage data to realize interframe and switches, the restructuring of picture and image quality post-processing algorithm, the most firmly Part realizes.What difficulty was maximum among these is exactly the integer process of floating type bore hole 3D screen and intercepts the process of certain precision floating point data The hardware of algorithm realize, and the error compensation hardware after processing for corresponding floating-point realizes the optimization of algorithm, specifically walks Rapid as follows:

1, to sub-pixel (x, y) calculating of accurate viewpoint value:

Here x represents input picture lateral coordinates position, and y represents input picture longitudinal coordinate position.According to eye 3D screen Parameter (grating slope angle, grating paster slant range), to input the every a line of picture each RGB data carry out viewpoint Calculate, the formula according to seeking viewpoint: N_(x,y)=M_out/P*V_viewCalculating viewpoint value, it must be a decimal, its fractional part Divide i.e. viewpoint weight to take precision 1/256, i.e. retain eight floating numbers；

This calculating comprises the complex calculation such as multiplication and division, and it is many that Hardware Implementation takies resource, but the inventive method only calculates Every first sub-pixel of a line, other overwhelming majority sub-pixel need not calculate, and direct result makes to realize by hardware exactly Bore hole 3D algorithm become to achieve with low cost hardware；In above formula, N is viewpoint value, M_outFor display screen physical resolution with Grating slope moves up and down the computational constant of relation, and P is grating physical parameter, V_viewNumber for viewpoint.

2, first sub-pixel calculated viewpoint value N of every a line is cast out process and obtain its integer part, i.e. N Deduct integer value N that the floating-point part of N obtains_Z；

3, the integer part of the N of the sub-pixel position of every a line is estimated, such as: if this picture is to use 8 to regard Point merges, then the mode according to { 1,2,3,4,5,6,7,8,1,2,3,4,5,6,7,8 ... } arranges estimation；

4, arranging the floating-point part of the N of the sub-pixel position of every a line, rule is by by first sub-pixel of this journey The floating-point part being calculated N is averaged arranged evenly；

5, the calculating to the floating-point viewpoint weight of the N of the sub-pixel position of every a line: open from second sub-pixel of every a line Begin estimation, floating-point increase by sub-pixel position and increments to zero, then the value of first sub-pixel from zero increase to this journey, as This circulation, until last sub-pixel position of this journey has been estimated；

6, the viewpoint value of estimation is compensated: by the N value of calculating in step 1 in addition to eight floating numbers retained, will be super Cross 8, need the floating-point partial value cast out to be defined as F, when specifically compensating, by the F of first sub-pixel according to in step D It is evenly distributed the mode backward mode with distribution, the sub-pixel floating-point weight of co-located is compensated, obtains final floating-point Part is N_F；

7, by formula N_Eventually=N_Z*N_F+(N_Z+1)*(1-N_F) calculate the viewpoint value that final upper screen multiple views merges.

Comparison software of the present invention calculates the method for viewpoint, obtains result and has error slightly, but combines regarding of human eye Feel feature, error control in the receptible scope of human eye, employing can hard-wired system architecture, improve the effect of algorithm Rate, allows the bore hole 3D displaying scheme cost of costliness be reduced significantly.

Claims

1. can hard-wired floating type viewpoint processing method in bore hole 3D display system, it is characterised in that comprise the following steps:

D, floating-point part to the N of the sub-pixel position of every a line are averaged arranged evenly and carry out floating-point weight estimation；

2. in bore hole 3D display system as claimed in claim 1 can hard-wired floating type viewpoint processing method, its feature It is, in step A, to the method for the viewpoint value calculating of first sub-pixel of every a line is:

According to seeking viewpoint formula: N=M_out/P*V_viewCalculate viewpoint value, the viewpoint value asked for is retained eight floating-points；Wherein, N For viewpoint value, M_outMove up and down the computational constant of relation for display screen physical resolution and grating slope, P is grating physics ginseng Number, V_viewNumber for viewpoint.

3. in bore hole 3D display system as claimed in claim 2 can hard-wired floating type viewpoint processing method, its feature It is, in step C, the integer part of the N of the sub-pixel position of every a line is arranged in order and enters according to the rule moved in circles Row estimation.

4. in bore hole 3D display system as claimed in claim 3 can hard-wired floating type viewpoint processing method, its feature Being, in step D, the method for described floating-point weight estimation is:

Starting estimation from second sub-pixel of every a line, floating-point is increased by sub-pixel position, and increments is to zero, then from zero It is incremented to the value of first sub-pixel of this journey, so circulates, until the floating-point weight estimation of last sub-pixel position of this journey Complete.

5. in bore hole 3D display system as claimed in claim 4 can hard-wired floating type viewpoint processing method, its feature Being, in step E, the described error bringing viewpoint estimation carries out the backward compensation with the regularity of distribution, specifically includes:

The floating-point part cast out is needed to be defined as F, by first in addition to eight floating numbers retained the viewpoint value asked in step A The sub-pixel floating-point of co-located, according to the mode of distribution same with the mode that the is evenly distributed backward in step D, is weighed by the F of individual sub-pixel Heavily compensating, obtaining final floating-point part is N_F。

6. in bore hole 3D display system as claimed in claim 5 can hard-wired floating type viewpoint processing method, its feature Being, in step F, the floating-point weight portion after the described viewpoint integer part according to estimation and compensation calculates final viewpoint value Method is: